Graded Contribution of Retinal Maturation to the Development of Oxygen-Induced Retinopathy in Rats

Abstract

purpose. Newborn rats exposed to hyperoxia during the first days of life have been shown to exhibit not only vasculopathy but also permanent changes in the structure and function of the retina. Given that the rat retina is immature at birth and that the maturation process continues until the opening of the eyes at 14 days of life, this study was conducted to investigate the susceptibility of the retina to oxygen toxicity as a function of the degree of retinal maturity reached at the time of oxygen exposure.

methods. Newborn rats were exposed to hyperoxia during selected postnatal day intervals. Scotopic electroretinograms were recorded at 30 and 60 days of age, and retinal histology was obtained at the end of the study.

results. There was a strong correlation between the duration of the hyperoxic event and the structural and functional consequences in the retina. However, the repercussions were significantly more profound when the exposure to oxygen occurred within the second week of life (6–14 days), compared with earlier (0–6 days) or later periods (14–28 days).

conclusions. The results strongly suggest that the structural and functional retinal changes secondary to postnatal hyperoxia are not only the direct consequence of exposure to high levels of oxygen (i.e., free radicals), but also are determined by the level of retinal maturity reached at the time of oxygen exposure. The results also indicate that the structural anomalies precede the functional impairments.

Retinopathy of prematurity (ROP) is the major eye disease of the premature newborn, especially those with a birth weight of 1000 g or those born before the 29th week of gestation. ^{1 2 3} Prematurely born infants are placed in a hyperoxic environment because of their unstable pulmonary status and relatively low fetal oxygen pressure. However, in response to the high concentration of oxygen, the blood vessels of the immature retina constrict and then are obliterated, stopping their normal maturational growth toward the peripheral retina, which then becomes avascular. ^{3 4 5} On return to a normoxic environment, the peripheral retina, which is subjected to relative hypoxia due to the absence of an adequate blood supply, grows new vessels ^{3 4 5} —a phenomenon known as neovascularization. This neovascularization can lead to a detachment of the retina ^{3 4} and, ultimately, to blindness. However, more recent evidence has revealed that even the milder forms of this retinal disorder can cause permanent functional sequelae. ^{6 7 8}  

Several animal models of oxygen-induced retinopathy (OIR), such as cats, dogs, mice, rats, and pigs, have been used to study ROP. ^{9 10 11 12 13 14} After exposure to hyperoxia, newborn pups not only have vasculopathy, which includes neovascularization, but also show permanent changes in retinal function due to failure in development of the outer plexiform layer (OPL). ¹⁵ Although it is well known that to generate OIR or ROP there must be a combination of an immature retina and hyperoxia, to our knowledge no one has examined the relationship between the degree of retinal immaturity and the severity of the retinopathy. For example, there is evidence that the rat retina exhibits an increased metabolic rate in the second to third week of life and that the retina undergoes significant maturation during this time. ^{16 17} Because increased metabolism is associated with increased leak of electrons from mitochondria, facilitating free radical generation, we therefore hypothesized that the retina of the rat is particularly susceptible to hyperoxia in the second week of life. Our data suggest that such is the case. 

Methods

Experiments were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the local animal care committee. Newborn Sprague–Dawley rat pups were placed in a controlled (12-hour light–dark cycle) environment. The illumination inside the cages varied between 20 and 30 candelas (cd)/m² depending on the animals’ positions relative to the ceiling lights—an intensity range well below that previously shown to yield retinal degeneration in albino rats. ¹⁸  

Pups were exposed daily to 80% O₂, interrupted by three 0.5-hour periods of 21% O₂ according to the following protocols: exposure from birth through day 3 (n= 7), 6 (n = 8), 9 (n = 7), 12 (n= 5), or 14 (n = 8) of life (i.e., 0–3, 0–6, 0–9, 0–12, or 0–14 days); or exposure from day 6 (n = 7), 9 (n = 7), or 12 (n = 5) through day 14 of life (i.e., 6–14, 9–14, or 12–14 days). Another group of rats (n= 6) was exposed from days 14 through 28. Age-matched control animals (n = 14) were raised simultaneously in normoxia. After the hyperoxic event, the rats were maintained in a normoxic environment up to a maximum age of 64 days, at which time the rats were killed. 

Electroretinography

The scotopic ERG was used to assess the status of the retina, because this measure was shown to yield a reliable estimate of the extent of OIR-induced functional deficits. ^{15 19 20 21} Because we were more interested in the permanent structural and functional deficits, the effect of postnatal hyperoxia was evaluated in rats aged 30 and 60 days. Under dim red light illumination, after a 12-hour period of dark-adaptation, the rats were anesthetized with an intramuscular injection of ketamine hydrochloride (80 mg/kg) and xylazine (6 mg/kg). They were then placed, lying on their sides, in an opaque recording chamber, in which the top part housed the rod desensitizing background light and a photostimulator (model PS22; Grass, Quincy, MA). ²² The cornea was anesthetized with proparacaine hydrochloride 0.5%, and the pupil was dilated with cyclopentolate hydrochloride 1%. The electroretinogram (ERG) was recorded with a DTL fiber electrode maintained on the cornea with a drop of 2% methylcellulose, which also prevented the desiccation of the cornea.²³ A 6-mm silver disc electrode (model E6GH; Grass) inserted in the mouth served as the reference electrode, and a platinum subdermal needle electrode (model E2; Grass) inserted in the tail served as the ground. Retinal potentials were amplified ×10,000 and recorded within a bandwidth of 1 to 1000 Hz with a analog preamplifier (model P511; Grass). Scotopic (rod-dominated) ERGs were evoked to flashes of white light spanning a 7.2-log-unit range (maximal intensity: 8 cd/m² per second as measured with a radiometer model IL 1700; International Light, Newburyport, MA), in approximately 0.3-log increments, for a total of 19 different ERG responses, each of which represented an average of two to five flashes (interstimulus interval, 10.24 seconds) depending on the intensity of stimulation. Recordings were performed with a data acquisition system (Acknowledge; model MP 100WS; Biopac, Goleta, CA). 

Retinal Histology

Histology was performed immediately after euthanasia by carbon dioxide. Samples of the retinas were embedded in Epon according to the technique previously reported. ¹⁵ Semithin (0.7 μm) sections of the central, nasal, and temporal retinas were stained with toluidine blue. Retinal layer thickness and cell counts were measured over a length of 780 μm. Because there was no evidence of a sectorial difference, the thickness of the OPL and the count of the horizontal cells represent the mean from the three sectors identified earlier (minimum, three measures per sector). Horizontal cells were identified by their size and pale staining with toluidine blue (compared with the dark staining of the nuclei from outer and inner nuclear layers). Three animals per regimen were studied. 

Data Analysis

The peak times and the amplitudes of the ERG components were measured according to the standard practice. ²⁴ The amplitude of the a-wave was measured from baseline to trough, whereas that of the b-wave was measured from the trough of the a-wave to the peak of the b-wave. Peak times were measured from flash onset to peak. Analysis of the a-wave was also performed according to the method previously described by Hood and Birch, ^{25 26 27} in which the rod a-wave amplitude data are fitted to the following equation, based on the Lamb and Pugh model ²⁸ :  

\[P_{3}\ (I,\ t){\cong}{\{}1-\mathrm{exp}{[}\mathrm{-}I{\cdot}S{\cdot}(t-t_{\mathrm{d}})^{2}{]}{\}}{\cdot}R_{\mathrm{m}(\mathrm{P}3)}\ \mathrm{for\ }t{>}t_{\mathrm{d}}\]

where P ₃ represents the sum of the individual rod responses and is a function of flash energy I and time t after the occurrence of a short flash. S is a sensitivity parameter that scales the intensity of the flash, R _m is the maximal response amplitude, and t _d represents a brief delay. Parameters R _m and S were obtained with commercial software (MatLab; MathWorks, Natick, MA). 

In addition, for each animal, the amplitude of the b-wave was plotted against the corresponding flash intensity to generate a scotopic luminance–response function curve. Sigmoidal dose–response regression curves (Prism 2.01 software; GraphPad, San Diego, CA) were chosen to fit our data, because they yielded the highest r ² values (see Figs. 2A : 0.99, 0.99; 2B: 1.00, 0.99; 2C: 0.99, 0.99; 2D: 0.99, 0.99). Analysis of individual data revealed that the intensity of stimulation necessary for the b-wave to reach saturation varied between −3.6 and −3.0 log of attenuation. Therefore, we arbitrarily identified the ERG evoked by the− 3.3 log unit of attenuation flash as that representing saturation of the scotopic b-wave. This value was used to calculate the rod V _max and retinal sensitivity (k) parameters, according to a method previously reported. ²⁹  

Statistical analysis, by which we examined the effect of the different oxygen regimens on the ERG, was performed with the use of two-way analysis of variance (ANOVA; regimens versus age, P < 0.05) followed by simple main effect. One-way ANOVA (P < 0.05) and Tukey’s honestly significant difference (HSD) test, as a post hoc pair-wise comparison, examined the effect of the different oxygen regimens on retinal cytoarchitecture parameters. The data are presented as mean ± SD. 

Results

Postnatal Hyperoxia and Retinal Electrophysiology

As shown in Figure 1 , progressively brighter flashes produced the expected gradual increase in the amplitude of both a- and b-waves, accompanied by a shortening in the timing of both waves. This feature was observed in the responses obtained from the normal rats (control) as well as those exposed to the different hyperoxic regimens (identified at the top of each panel as 0–6: exposed from day 0 to day 6; 0–9: exposed from day 0 to day 9; and so on). As the duration of oxygen exposure increased, there was a gradual reduction in the amplitude of the b-wave, whereas the amplitude of the a-wave was minimally affected. This is best illustrated with the results shown at the bottom right panel of Figure 1 , showing representative ERG responses evoked to three different flash intensities (−0.3, −3.3 and −6.6 log units of attenuation) obtained after exposure to the four different regimens of hyperoxia (exposed between days 0 and 6, 0 and 9, 0 and 12, and 0 and 14) are compared with the normal (control) responses. 

When hyperoxia was initiated at birth, there was a dose–response correlation between the duration of hyperoxia and the amplitude of the resultant rod b-wave V _max (Fig. 2A ), which can be demonstrated in responses obtained from the 30- and 60-day-old animals. Although oxygen exposure during the first 6 days of life did not affect the amplitude of the rod b-wave V _max measured at 30 days of age, there was a near linear decline in b-wave amplitude when the data from the remaining three O₂ exposure intervals (0–9, 0–12, and 0–14 days) were considered. The mean reduction in rod b-wave V _max amplitude was 25%, 55%, and 70% for the three intervals, ;T1>respectively (Fig. 2A , Table 1 ). Although larger in most instances, the amplitudes of the rod b-wave V _max measured in rats aged 30 days were not significantly different from those measured at the age of 60 days. 

A similar dose–response correlation between the duration of hyperoxia and the amplitude of the rod b-wave V _max was also observed when oxygen exposure was initiated at a later age (Fig. 2B) . The amplitude of the rod b-wave V _max measured from 30-day-old rats exposed during the 6- to 14- and 9- to 14-day intervals reached approximately 30% of control. It is noteworthy that the latter value was not significantly different from that reached after exposure during the 0- to 14-day interval, probably because, as shown in Figure 2A , exposure within the 0- to 6-day interval had a minimal impact on the ERG. In comparison, oxygen exposure during the 12- to 14-day interval reduced the amplitude of the rod b-wave V _max to an average 70% of control amplitude (Fig. 2B , Table 1 ). A comparable effect was also observed in the 60-day-old rats (Fig. 2B , Table 1 ). However, a similar 30% attenuation in rod V _max amplitude was also obtained after oxygen exposure during the 14- to 28-day interval (Table 1) . 

In contrast, the different oxygen exposure regimens had no significant impact on the peak time of the rod V _max b-wave, irrespective of age, except for the 0- to 14- and 14- to 28-day regimens, which showed significantly longer peak times at 30 days of age. Furthermore, whereas in the 30-day-old rats none of the posthyperoxia retinal sensitivity measurements (k) was significantly different from controls, some k values measured in the 60-day-old rats suggest an increase in retinal sensitivity compared with normal animals. The regimens 0 to 14 and 6 to 14 days yielded k parameters significantly higher than control levels, the latter also being different from the results obtained in the 30-day-old rats (Table 1) . 

As shown at Table 1 , the amplitude of the a-wave (when measured from baseline to trough) did not demonstrate a similar dose–response correlation. The amplitude of the a-wave was significantly smaller than normal only for the 0- to 14- (19% attenuation), 9- to 14- (19% of attenuation), 12- to 14- (20% of attenuation), and 14- to 28- (31% of attenuation) day intervals. That there was no dose–effect relationship was further confirmed with the analysis of the a-wave performed according to the equation given in the Methods section. As shown at Table 2 , the amplitude of P ₃ (R _m), measured in 30-day-old rats, was significantly smaller than normal (25% attenuation) after the 0- to 14-, 9- to 14-, and 12- to 14-day intervals, whereas an exposure within the 6- to 14-day interval did not significantly alter the P ₃ component. In comparison, measurements obtained after exposure within the 14- to 28-day interval resulted in a 30% reduction in the amplitude of P ₃ irrespective of the age of the rats. Similarly, the S parameter increased by approximately 0.1 log in the 30-day-old rats and by 0.2 log in the 60-day-old rats—increments that were significantly different from normal animals (at both ages) after exposures within the 0- to 14- and 0- to 9-day intervals. Finally, unlike b-wave measurements, it is interesting to note that, irrespective of the exposure regimen, a-wave amplitude measurements were always smaller in the responses taken at 60 days. However, the latter amplitude reduction, which was also seen in the normal rats, were not accompanied by a significant modification in the photoreceptor sensitivity (S) parameter. 

Postnatal Hyperoxia and Retinal Cytoarchitecture

A progressive increase in the duration of oxygen exposure caused a gradual thinning of the OPL (Fig. 3) . Also observed were a reduction in horizontal cell count, a loss of the regular stacking of the nuclei of the ONL, and the appearance (at the same retinal depth as that of the OPL) of vacuolization. These histopathologic features were best evidenced after exposure within the 0- to 12- and 0- to 14-day intervals. 

As shown at Figure 2C (and Table 1 ), oxygen exposure during the first 6 and 9 days of life resulted in significant reductions in OPL thickness of 30% and 50%, respectively, whereas exposures during the first 12 and 14 days of life resulted in an equivalent 80% reduction. In contrast, the relationship between the horizontal cell counts and oxygen exposure was not as well defined. The cellular count obtained after exposure within the 0- to 6-day interval was not significantly different from control counts, and counts performed after the remaining exposures (0–9, 0–12, 0–14 days) were not significantly different from each other, although they were significantly lower (mean 80% reduction) than control counts. When initiated later in life, oxygen exposure reduced OPL thickness to 20% of control thickness during the 6- to 14-day interval, and to 35% during the 9- to 14- and 12- to 14-day intervals. These values were neither significantly different from each other nor from those obtained after the 0- to 14-day exposure, but all were significantly lower than control thickness. Similar results were also obtained with the horizontal cell counts. All three experimental rat groups showed a reduction in horizontal cell counts to values less than 40% of control counts (Fig. 2D , Table 1 ). In comparison, the thickness of the OPL resulting from the 14- to 28-day exposure was not significantly different from the control thickness, whereas the number of horizontal cells was reduced to 50% of normal (Fig. 4 , Table 1 ). 

Our results also indicate that for exposure regimens of equivalent duration (within the first 14 days of life), those initiated earlier were less detrimental to the retinal structure and function (b-wave measurements) than those that began later. For example, an 8-day exposure from postnatal days 6 to 14 reduced the rod V _max, thickness of the OPL, and horizontal cell count more than a 9-day exposure from birth to postnatal day 9. Similarly, a 5-day exposure from postnatal days 9 to 14 caused more damage to the retina than a 6-day exposure from birth to postnatal day 6. In fact, oxygen exposure for 5 days (9–14) was more detrimental to the retinal structure and physiology than a 9-day exposure beginning at birth (0–9). Comparing two nonoverlapping regimens of similar duration (0–3 and 12–14 days) revealed that there was no significant change in retinal function after the 0- to 3-day regimen, whereas there was a significant 30% reduction in b-wave amplitudes after the 12- to 14-day regimen. Finally, a 14-day exposure initiated at an older age (days 14–28), produced functional changes comparable (if not identical) with those seen after a shorter, 2-day exposure, begun earlier (postnatal days 12–14). 

Discussion

We previously reported ¹⁵ that postnatal hyperoxia, during the first 14 days of life, causes permanent loss of the OPL, severe reduction in the number of horizontal cells, and a permanent deterioration of the retinal function, as determined with the ERG (photopic, scotopic, and oscillatory potentials). The present study not only confirms our initial observation but further demonstrates that the structural and functional consequences of postnatal hyperoxia exhibit a duration–effect relationship that is dependent on the degree of maturity of the retina. Although other studies have suggested that the duration of oxygen exposure could contribute to the vascular anomalies typical of OIR (or ROP), ^{30 31 32 33 34 35} to our knowledge this study is the first to demonstrate that the degree of retinal maturity reached at the time of oxygen exposure also plays an important role in the structural and functional consequences of postnatal hyperoxia. 

The gradual reduction in rod b-wave V _max amplitude and in OPL thickness as a function of length of oxygen exposure (within the first 14 days of life) suggests that the severity of the structural and functional deficits is solely due to the amount of oxygen that the young rats were exposed to. However, if the retina is exposed for the same duration but later in life (e.g., the 14- to 28-day interval), this relationship no longer holds, because the rod b-wave V _max is only reduced to 70% of control value, whereas the thickness of the OPL remains unchanged. Interestingly, however, the horizontal cells always appear to be highly susceptible to hyperoxia, irrespective of the exposure interval, although the most vulnerability occurs within the second week of life (6- to 12-day interval). 

In comparison, the different oxygen exposure regimens did not yield a similar dose–effect correlation to a-wave parameters. In fact, almost all exposure regimens that included at least days 6 to 9, for the S parameter, and days 12 to 14, for the amplitude parameters, resulted in significantly abnormal measurements (Table 1) . This was most pronounced for measurements obtained after hyperoxia within the 14- to 28-day interval, during which the a-wave showed its most significant reduction in amplitude (Table 2) . If we interpret this reduction in amplitude of the a-wave as indicative of a loss (structural or functional) of photoreceptors, then our finding would oppose that of Maslim et al. ³⁶ who showed that exposure to hyperoxia (70%–75%) between postnatal days 15 and 22 not only retarded the normal elimination of the photoreceptors of normal albino rats but also lengthened the survival of the degenerating photoreceptors of the Royal College of Surgeons (RCS) rats, suggesting that oxygen availability during that critical period was crucial to the development of the photoreceptors. In the same study, the authors showed that hypoxia was detrimental to the survival of the photoreceptors, especially after postnatal day 21. Our different hyperoxic protocols may untangle the apparent discrepancy. Whereas in protocol in Maslim et al.³⁶ the animals were subjected to a constant level of hyperoxia throughout the exposure, in our study, the hyperoxic regimen was interrupted by brief periods of normoxia. These periods of normoxia were previously suggested to give rise to short periods of relative hypoxia, partly due to the vasoconstrictive effect of the preceding hyperoxic event. ³¹ It could be that even these short periods of relative hypoxia, which appear to have very little impact on the function of the photoreceptors when occurring earlier in life, become significantly more harmful later in development when oxygen availability is so critical to their survival. ³⁶  

The relative immunity of the photoreceptors to hyperoxia during the first week of life or so could be explained by the fact that the outer segment of the photoreceptors only reach maturity during the second week of life. ³⁷ In contrast, the dramatic reduction in b-wave amplitude strongly suggests that the most important, and more permanent, sequelae of OIR occur in the more inner part of the retina. We believe that our results are in line with those reported by Reynaud et al., ²⁰ who examined the retinal function of infant rats aged 13 and 18 days that had been exposed to postnatal hyperoxia for the first 11 days of life. They showed that whereas a-wave parameters (amplitude and sensitivity) improved with age, b-wave measurements obtained at 18 days showed a significant deterioration compared with those obtained at 13 days of age, thus confirming the increased and more permanent susceptibility of postreceptoral elements to postnatal hyperoxia. 

Free radicals generated by lipid peroxidation may be an important factor in the pathophysiology of OIR (or human ROP), because postnatal hyperoxia in kittens causes a significant reduction in retinal superoxide dismutase, ³⁸ and vitamin E (free radical scavenger) supplementation significantly attenuates the severity of OIR in rats. ³⁹ However, our results indicate that a mere 2-day exposure later in life (12–14 days) alters the retinal structure and function to a similar extent as a 9-day exposure started at birth. Furthermore, exposure of a similar duration early in life (0–3 days) did not change retinal function at all. This would suggest that free radicals may play only a limited role in the pathogenesis of OIR (and ROP), because oxygen exposure of a longer duration did not always result in a more severe form of OIR. 

The degree of severity could, for instance, be related to the order of maturation of the different retinal elements. In mice, the horizontal cells are among the first retinal cells (with the amacrines, cones and ganglion cells) to be born at approximately embryonic day 14, whereas the rods appear at birth and the bipolar and Müller cells at postnatal days 3 to 4. ⁴⁰ The formation of the OPL begins at approximately postnatal day 5, presumably as a result of the lateral growth of the horizontal cells. ^{37 40} From days 6 to 9, the OPL has established some connections with the photoreceptors but not with the inner nuclear layer (INL); therefore, no electroretinographic activity can be recorded. ^{35 40 41 42 43} The development of the INL follows, and synaptic connections are instituted at approximately 10 to 12 days, when electroretinographic responses are obtained. On day 12, the retina already has an adult appearance, although fine-tuning continues for another 2 to 3 weeks. ³⁷ Consequently, it appears that exposing the immature retina to hyperoxia, can prevent the formation of synapses in the OPL and, most probably, the INL and/or outer nuclear layer (ONL), as well. The disappearance of the horizontal cells may result from a hypersensitivity to oxygen or the consequence of some degenerative process triggered by their failure to establish proper synaptic connections. Both pathophysiological processes would yield a thinner OPL. The abruptness of the slopes (Figs. 2C 2D) describing the changes in horizontal cell count with oxygen exposure instead suggests the existence of a critical period of oxygen vulnerability of the horizontal cells (between postnatal days 6 and 12), as opposed to a degeneration in which a more gradual reduction would be expected. Consequently, the synaptic exchange between the a-wave generators (photoreceptor outer segments) and the b-wave generators (postreceptoral retinal elements, such as bipolar and Müller cells) would be seriously compromised, resulting in a significantly attenuated b-wave. Similar findings were observed in transgenic mice expressing simian virus 40 T antigen, which induces a progressive degeneration of OPL and horizontal cells ⁴⁴ associated with a normal a-wave and a markedly attenuated b-wave. ⁴⁵  

The exact pathophysiological mechanisms at the origin of OIR or ROP remain to be fully understood. Although there is strong evidence that postnatal hyperoxia plays a significant role, our demonstration of a graded, maturation-linked susceptibility suggests the possibility of other contributing factors. For example, due to our study design, which entailed an initial ERG recording at 30 days of age (that is, 16 days after the cessation of oxygen exposure), we could not evaluate the possible contribution of the relative retinal hypoxia, which is known to immediately follow the cessation of the hyperoxic regimen. It is well documented that, in the normal maturing retina, hypoxia serves as a one of the developmental cues for the photoreceptors and retinal vessels. ^{36 46 47} In the latter case, it has been postulated that the local oxygen tension regulates the activation of two antagonistic growth factors, vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF), in which the former activates vasculogenesis, thus permitting the normal maturation of the retinal blood vessels, and the latter suppresses it, once retinal maturation ends in an adequate vascular coverage. ⁴⁶ In contrast, in retinal diseases, such as OIR (or ROP), the relative hypoxia that follows the hyperoxic event will induce a disorganized formation of new vessels (neovascularization) that are the origin of the most severe consequences of this retinal disorder. ^{30 48 49}  

However, given that we did not measure the degree of neovascularization that resulted from exposure to the different regimens, we cannot use this other feature of OIR to quantify the severity of the retinopathy that was generated. Similarly, we cannot rule out the possibility that it is the hypoxia (more than the hyperoxia) that induced the death of the horizontal cells, in a way similar to the death of the photoreceptors of the RCS rats, which is also precipitated by hypoxia, despite the fact that the horizontal cells are suggested to play a protective role against hypoxia. ^{50 51} However, despite this limitation, we were able to clearly demonstrate that during postnatal retinal development, there is a temporal window of transient increase in oxygen vulnerability: hyperoxic events occurring immediately after birth exerting a lesser impact on the retinal structure and function than when a relatively more mature retina is subjected to the same insult. 

It is interesting to note, as shown in Figure 5 , that the ;F5>temporal window of oxygen vulnerability that we identified, with our analysis of the ERG and retinal cytoarchitecture, exactly matches that previously evidenced by Graymore, ^{16 17} who showed that the rat’s retina doubles its oxygen consumption during the second week of life. Postnatal hyperoxia, in preventing the normal growth of the retinal blood vessels thus seriously compromises this transient increase in the metabolic demand of the retina and consequently significantly impairs the normal development of the retinal tissue. This leads, as we have shown, to permanent structural and functional damage. In contrast, as we have also shown, oxygen exposure at a later age has a lesser impact and its detrimental effect mostly concentrates in the photoreceptors. In that respect, it is interesting to note that previous studies have shown that the onset of the clinical signs of human ROP occurs at a postnatal age that is dependent on the state of maturation at birth. The onset of ROP occurs at a later postnatal age in the more premature neonates than in the more mature ones. ⁵² This finding supports our concept of a critical period of oxygen vulnerability for the retinal cytoarchitecture and function, similar to that previously reported for retinal blood vessels. ⁵³  

 Supported by grants-in-aid from the McGill University-Montreal Children’s Hospital Research Institute, by the Medical Research Council of Canada (Grant MT-12153 and MT-13383) and by the FCAR.

 Submitted for publication June 23, 2000; revised November 16, 2000; accepted November 30, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Representative scotopic ERG responses obtained at age 30 days from a control rat and a rat subjected to postnatal hyperoxia during the periods indicated at the top of each panel. In each, ERG responses were evoked to progressively brighter flashes (from bottom to top) as indicated at the left of tracings (in log-units of attenuation). In the bottom right panel, results obtained after the different oxygen exposures are compared with normal responses evoked at three different flash intensities. Horizontal and vertical calibrations: 20 msec and 500 μV, respectively; positivity upward. All tracings include a 20-msec prestimulus baseline. Vertical arrows: flash onset.

Figure 2.

 

The impact of oxygen exposure on the V _max (A, B), thickness (in micrometers) of the OPL, and horizontal cell count (C, D) in absolute values. For data illustrated in (A) and (C), oxygen exposure started at birth and ended at different postnatal ages, as indicated with the first ordinate. Similarly, for data illustrated in (B) and (D), oxygen exposure started at different postnatal ages and ended at day 14 as indicated with the first ordinate. A second ordinate indicates the total number of days of oxygen exposure. The latter two measures were obtained from 60-day-old rats. Data points represent mean ± SD.

Table 1.

 

Summary of Amplitude Measurements, Peak Times of b-Wave Rod V_max, Retinal Sensitivity, OPL Thickness, and Horizontal Cell Counts

Parameters	Regimens
		Control (0)	0–3 (3)	0–6 (6)	0–9 (9)	0–12 (12)	0–14 (14)	6–14 (8)	9–14 (5)	12–14 (2)	14–28 (14)
30 days
Rod Vmax (μV)	676.4 ± 98.6	660.7 ± 65.2, †	649.0 ± 65.2, †	501.4 ± 81.8* , †	292 ± 137.3*	193.1 ± 135.9*	226.0 ± 75.9*	292.4 ± 127.9*	496.9 ± 74.4* , †	496.8 ± 37.4* , †
Rod Vmax (msec)	64.2 ± 4.2	63.5 ± 7.0, †	66.4 ± 5.1	66.7 ± 4.4	66.3 ± 5.2	70.0 ± 10.4*	65.6 ± 9.7	68.1 ± 8.6	66.1 ± 4.6	71.5 ± 6.5
k (log of att)	−5.60 ± 0.25	−5.71 ± 0.11	−5.65 ± 0.10	−5.71 ± 0.10	−5.87 ± 0.49	−5.78 ± 0.48	−5.61 ± 0.42	−5.72 ± 0.17	−5.60 ± 0.24	−5.81 ± 0.05
60 days
Rod Vmax (μV)	609.7 ± 73.6	NP	587.8 ± 95.4, †	416.8 ± 104* , †	303.2 ± 105.8*	224.8 ± 131.5*	201.9 ± 74.0*	347.7 ± 147.1* , †	429.8 ± 125.4* , †	445.7 ± 85.8* , †
Rod Vmax (msec)	62.7 ± 3.3	NP	60.6 ± 5.1	65.9 ± 4.6	62.4 ± 4.4	64.4 ± 6.6	63.4 ± 7.4	63.9 ± 3.1	64.3 ± 1.8	65.3 ± 4.8
k (log of att)	−5.73 ± 0.10	NP	−5.71 ± 0.21, †	−5.72 ± 0.11, †	−5.84 ± 0.28	−6.06 ± 0.61*	−5.91 ± 0.57* , ‡	−5.67 ± 0.64, †	−5.57 ± 0.29, †	−5.64 ± 0.07, †
OPL (μm)	13.10 ± 0.12	NP	9.80 ± 0.36* , †	6.60 ± 0.30* , †	3.03 ± 1.56*	2.77 ± 1.46*	2.27 ± 1.21*	4.47 ± 1.46*	4.43 ± 0.85*	11.50 ± 0.26, †
Cell counts	5.77 ± 0.64	NP	4.40 ± 0.53, †	0.87 ± 0.51*	0.87 ± 0.12*	1.03 ± 0.75*	0.67 ± 0.15*	0.17 ± 0.29*	2.33 ± 1.16*	2.83 ± 0.17* , †

 Measurements were obtained after oxygen exposure during predetermined intervals as indicated at the top of each column where the number within parentheses indicates the duration of exposure in days. b-Wave stimulus was −0.3 log unit of attenuation. Data are means ± SD. NP, not performed.

*  Significantly different from Control.

†  Significantly different from 0–14 days.

‡  Significantly different from 30 days.

Table 2.

 

Summary of a-Wave Amplitude, Peak Time Measurements Taken at Maximal Response, and Photoreceptor Sensitivity S

Parameters	Regimens
		Control (0)	0–3 (3)	0–6 (6)	0–9 (9)	0–12 (12)	0–14 (14)	6–14 (8)	9–14 (5)	12–14 (2)	14–28 (14)
30 days
a-Wave (μV)	476.6 ± 70.9	425.3 ± 44.3	505.6 ± 52.0†	450.8 ± 56.0†	432.6 ± 34.3	386.0 ± 55.1*	476.9 ± 46.4†	387.9 ± 93.9*	381.9 ± 58.4*	327.1 ± 47.7*†
a-Wave (msec)	9.96 ± 0.72	9.51 ± 0.41†	9.93 ± 0.50†	9.94 ± 0.75†	10.20 ± 0.91	10.55 ± 0.53*	10.03 ± 0.56	9.89 ± 0.46†	10.12 ± 0.34	9.43 ± 0.46†
Rm (μV)	478.4 ± 75.9	430.3 ± 42.3	500.1 ± 54.8†	445.4 ± 57.9	426.5 ± 37.5	388.4 ± 60.0*	479.5 ± 48.7†	389.8 ± 81.2*	386.6 ± 55.0*	323.5 ± 46.5*†
S (log)	5.43 ± 0.1	5.41 ± 0.07†	5.43 ± 0.05†	5.34 ± 0.12*	5.42 ± 0.12†	5.30 ± 0.05*	5.36 ± 0.09	5.41 ± 0.08†	5.39 ± 0.05	5.34 ± 0.09
60 days
a-Wave (μV)	383.1 ± 53.7‡	NP	421.2 ± 46.6†‡	342.2 ± 60.6‡	343.3 ± 43.1‡	313.5 ± 45.5*‡	321.8 ± 65.7*‡	308.1 ± 35.8*	322.6 ± 52.3*	246.8 ± 15.8*†‡
a-Wave (msec)	9.67 ± 0.52	NP	9.58 ± 0.36†	9.83 ± 0.90	10.20 ± 0.71	10.34 ± 0.53*	9.58 ± 0.48†	9.77 ± 0.34	9.92 ± 0.52	9.60 ± 0.60†
Rm (μV)	366.2 ± 55.2‡	NP	409.0 ± 45.3†‡	330.8 ± 60.2‡	344.4 ± 45.4‡	314.2 ± 47.3*‡	315.9 ± 65.9*‡	299.7 ± 33.1*‡	311.3 ± 44.8‡	243.3 ± 18.5*†‡
S (log)	5.46 ± 0.08	NP	5.44 ± 0.07†	5.35 ± 0.12*†	5.33 ± 0.08*	5.23 ± 0.09*	5.40 ± 0.09†	5.39 ± 0.06†	5.33 ± 0.12*	5.30 ± 0.08*

 Measurements were obtained after oxygen exposure during predetermined intervals, as indicated at the top of each column, where the number within parentheses indicates the duration of exposure in days. Data are means ± SD. Description of symbols used for statistical differences is as in Table 1 .

Figure 3.

 

Top: Photomicrograph of histologic sections of the central retinas from 60-day-old control rats and rats exposed to hyperoxia during 0 to 6, 0 to 9, 0 to 12, and 0 to 14 days of life as indicated at the top of each section. Bottom: Magnification of the sections shown at the top to better illustrate the absent OPL. R, photoreceptor layer; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar, 10 μm.

Figure 4.

 

Photomicrograph comparing the central retina of a control rat (left) and that of a rat exposed to postnatal hyperoxia (right) from 14 to 28 days of age. Data are presented as in Figure 3 .

Figure 5.

 

Oxygen during the development of the rat retina. Age range between birth and adult is divided by days. The diagram compares the different postnatal windows of oxygen susceptibility reported in previous studies for the retinal vasculature and photoreceptors. These are compared with that evidenced in the present study for the OPL and ERG. These windows are compared with the window of peak oxygen consumption originally demonstrated by Graymore. ^{16 28}

The authors thank Sherif Shady for sharing his knowledge on a-wave modeling and Éric Simard for the experienced and dedicated care given to the animals.